The acquisition of semantic information from a scene with the help of video sensorics is a challenging task. The VID department is involved in the situation analysis in various areas. These include, inter alia, scenarios from the following discourses:
Based on person detectors and trackers, the behavior of groups of persons is analyzed in order to uncover any abnormality. Statistics can be created for documentation, which in turn can be the basis for subsequent evaluations.
In the case of road transport, it is particularly necessary to model the transport users and their intention in a suitable manner. For this purpose, the knowledge of context information, such as rules for traffic rules, traffic lights, signs, etc., is also important. Based on this, driving maneuvers can be derived automatically.
Left: Identification of a group of persons and tracking of individual persons for the detection of suspiciousness. Right: detection of a surrender in a human race.
Evaluation of a soccer match. The ball, the referee and the player are each tracked separately with separate split & merge treatment for intersecting trajectories (right). Statistics and parameters can be determined using trajectories.
Recognition of the situation in the area of driver assistance. Left: Object detection and prediction of the impact time and location. Right: Determination if the scene parameters in bad weather confitions and adapted procedures for the detection and tracking if traffic.